Update the multistart optimization tutorial in ensemble.md

[francis-biocurie]

Checklist

• Appropriate tests were added
• Any code changes were done in a way that does not break public API
• All documentation related to code changes were updated
• The new code follows the
  contributor guidelines, in particular the SciML Style Guide and
  COLPRAC.
• Any new documentation only uses public API

Additional context

I made the two following changes:

1. The i argument was not used, so the initial guesses were all identical.
2. I created two seperated markdown blocks, to be able to see the objective value of the first non-multistart optimization in the online documentation.

Also: maybe we should explain what is the repeat argument here ? I don't understand its role. I'm not able to find any explanation about this argument in the SciML documentation.

[ChrisRackauckas]

Also: maybe we should explain what is the repeat argument here ? I don't understand its role. I'm not able to find any explanation about this argument in the SciML documentation.

If the solution was rejected, then repeat=true so you know it's a repeat solve. It's something that is mostly useful for stochastic algorithms, like SDEs, which might fail on some noise instantiations. Stochastic optimizers could be a use case here: if you have a stochastic optimizer diverge, you can reject that from your ensemble and have it just run those again.